Most of the nine genera and 38 species in the family Pythonidae occur in south-east Asia and Australasia, but the genus Python is distinctive in that it also occurs in sub-Saharan Africa. There is a large distribution gap for Python between Africa and south-east Asia, but fossil evidence suggests Python once occurred in this gap, as well as in Europe until the mid-Miocene. However, because all African species have not previously been included in any previously published phylogeny, their monophyly has not been established. Further, it has been suggested that the genus Python may have an Asian origin, and this scenario would require the African species to be monophyletic. Otherwise, multiple independent dispersals into Africa would be required to explain current day biogeographic patterns. To test these competing hypotheses, a dated phylogeny was constructed based on one nuclear and two mitochondrial genes and biogeographic scenarios evaluated using ancestral range reconstruction. In addition, fossil evidence was appraised as supporting evidence for their paleo-distribution but also to evaluate the hypothesis that large pythons (e.g., > 6 m in body length) are restricted to warmer climatic zones, and we suggest this influenced their paleo- and present-day distributions. The dated phylogeny indicates monophyly for the African species, diverging from the Asian species approximately 33 Mya. Ancestral area reconstruction supports an Asian origin for the genus, with a single dispersal event to Africa after which species diversified across the continent. We find no support for multiple dispersal events into Africa. Accounting for the distribution of fossils, it appears that Python, including large-bodied species, were once more widespread, but ranges contracted in response to global cooling since the mid-Miocene, and they are now excluded from temperate (i.e., Europe, Iranian plateau) as well as hyper-arid areas (i.e., Arabian, Saharan and Namib deserts).

Africa, adaptive radiation, biogeography, Miocene cooling, Pythonidae, range limits, reptiles, snakes

Snakes are a diverse clade of squamates that arose around 170 million years ago (Zheng and Wiens 2016; Title et al. 2024). With more than 4,000 species, snakes are adapted to aquatic, fossorial, terrestrial and arboreal habitats and occur on every continent except Antarctica. They are divided into two main Infraorders, the Scolecophidia (fossorial thread and blind snakes), and the Alethinophidia (all other snakes). The python family (Pythonidae) consists of nine genera composed of 38 extant species and several extinct species (Uetz et al. 2025) and diverged from other snake lineages early in the evolution of the Alethinophidia, some 60–90 Mya (Pyron et al. 2013; Zheng and Wiens 2016; Zaher et al. 2019). Extant species of Pythonidae occur across southern Asia, Australia, Oceania and a large portion of Africa (Torr 2000; O’Shea 2007; Uetz et al. 2025; Figs 1, 2), while extinct species, and locally extinct populations of extant species, formerly occurred in Europe, western Asia, north Africa and southern Africa (Römer 1870; Rage 1976; Szyndlar and Rage 2003). Amongst snakes, pythons are known for their unique morphological characteristics and ecology (O’Shea 2007; Esquerré and Keogh 2016; Leal and Cohn 2016; Esquerré et al. 2017, 2020). Furthermore, they show extreme range in body size, with species ranging in adult body length from 0.7 m to 9 m (O’Shea 2007; Esquerré et al. 2017). Their ecological and phenotypic diversity is attributed to adaptive radiations into different habitats, but they also show convergence when species occupy similar ecological niches (Esquerré and Keogh 2016; Esquerré et al. 2017, 2020).

Within the Pythonidae, the genus Python is the most geographically widespread, occurring in four biogeographic regions – the Oriental, Oceanian, Australian and Afrotropical, although they are absent from the island of Madagascar. There is a large gap in the occurrence of Python across northern Africa, the Saharo-Arabian-Levant region and Iranian plateau (Figs 1, 2). However, fossil evidence dated to the early and mid-Miocene (20–12 Mya) shows that, historically, some species of Python occurred in this gap as well as along the western seaboard of Namibia (Thomas 1982; Szyndlar and Rage 2003) where they currently do not occur (Fig. 2, Suppl. material 1: table S1). Together with Miocene fossils from central and south-eastern Europe (Szyndlar and Rage 2003) and Pakistan (Hoffstetter 1964; Head 2000), there is evidence that Python was more widespread up until the mid-Miocene and occurred across much of Eurasia (Fig. 2, Suppl. material 1: table S1). Notably, there are no known python fossils from Madagascar, although fossil evidence in general from that island is rare (Krause et al. 2006).

Python females actively incubate and protect their eggs, and it is thought that ranges are limited to regions where environmental temperatures are suitable for successful incubation (Lillywhite 1987; Alexander 2007, 2018). Of the ten Python species, four have giant body sizes, can exceed 5 m (P. natalensis, P. sebae, P. molurus and P. bivittatus) and are restricted to the tropics and sub-tropics, supporting the hypothesis that large pythons are restricted to warmer climates (Ivanov and Böhme 2011). It is therefore plausible that long-term environmental cooling initiated in the Eocene, intensifying in the Miocene (Zachos et al. 2001a, b, 2008; Westerhold et al. 2020) has played a role in range contractions, vicariance and extinctions of Python species and populations.

Most Asian Python species have relatively small geographic ranges, apart from the two widespread, giant-bodied species (Fig. 2), P. bivittatus (Burmese Python) and P. molurus (Indian Rock Python). On average, African Python species have more extensive geographic distributions than most Asian species (Fig. 2). The large bodied (maximum recorded body length ~6 m) P. natalensis (Southern African Python) and P. sebae (Northern African Python) are the most widespread African species, with P. natalensis occurring across most of the southern parts of Africa and P. sebae in East and West Africa (Broadley 1984). These two species are morphologically similar (Fig. 3) and occur primarily in mesic to arid savanna, although P. sebae may extend into forest in parts of its range. Python sebae and P. natalensis appear to be sympatric in parts of East Africa (Fig. 2; Broadley 1999; Spawls et al. 2018) and parapatric or sympatric in other areas (Fig. 2). Their morphological and colouration similarities led to P. natalensis first being classified as a subspecies of P. sebae (Broadley 1984), although P. natalensis was later elevated to a full species (Broadley 1999). In contrast, P. anchietae (Anchieta’s Dwarf Python) and P. regius (Ball Python) are small- bodied (maximum body length < 1.8 m) and have smaller distributions than the large-bodied species (Fig. 2). Python anchietae occurs mainly in arid environments, extending into arid savanna in southern Angola (Branch 2018) and central and northern Namibia (Marques et al. 2018), while P. regius occurs in forest and mesic savanna extending from Central to West Africa (Spawls et al. 2018).

The present-day disjunct distribution of the genus, as well as their absence from Madagascar, is intriguing in terms of biogeography. Indeed, several competing hypotheses regarding the origin of the genus Python and family Pythonidae have been proposed, but these hypotheses have not been tested by means of comprehensive datasets, and none have addressed the significant gap in their distribution, nor their absence from Madagascar. For example, Reynolds et al. (2014) concluded that Python regius (from Africa) is sister to all other Python species (African and Asian species), suggesting a possible African origin for the Pythonidae. In a subsequent study, Esquerré et al. (2020), also recorded a basal divergence of P. regius from the rest of the genus dated to approximately 32 Mya, but it should be noted that P. regius was the only African species included in that phylogeny. An Asian origin for the entire family Pythonidae with subsequent dispersal into Africa and southwards into Australasia has been suggested based on the high species richness and morphological diversity in Austral-Asia (Branch and Erasmus 1984; Esquerré et al. 2020). This has been supported by a recent phylogenomic analysis for the family (Esquerré et al. 2022), although this phylogeny included only single individuals of three African species. We tested these prevailing hypotheses by generating a near-complete, dated phylogeny and historical biogeographical assessment of Python in Africa, with complete sampling for African species. This allowed us to specifically test for monophyly of the African species, the validity of P. natalensis as a full species, the geographic origin of the genus, and to explore possible scenarios as to their absence from Madagascar.

Figure 1. 

Biogeographic regions (from Holt et al. 2013) with the number of Pythonidae species that occur in each area. Numbers in brackets show the number of species occurring in each region for the entire family and for the genus Python (family Pythonidae / genus Python).

Figure 2. 

Current day distribution of the ten species in the genus Python and the locations of python fossils. Fossils are denoted for the early Miocene (black skulls) and the mid-Miocene (white skulls), with extinct species indicated by the red interdictory circles. All other fossils represent extant species, or those unidentified at the species level. DNA samples of African species included in the phylogenetic analyses (Table 1) are shown by coloured circles. For samples that lacked coordinates, the centre point of the country of collection (or other locality descriptor) was used for mapping. Pet trade samples were not mapped. Distribution polygons from https://www.iucnredlist.org.

Figure 3. 

Comparison between Python sebae (top) and Python natalensis (bottom) for colouration and scalation.

Thirty Python samples were included in the ingroup dataset, representing eight of the 10 species in the genus (Table 1; no comparative sequence data available on GenBank for P. breitensteini from Borneo or P. kyaiktiyo from Myanmar). Twenty of these were sequenced for this study, with the remainder of the data downloaded from GenBank (Table 1). In addition, sequence data from five individuals representing four outgroup species from other genera in Pythonidae were also downloaded from GenBank (Table 1). Some samples were composed of chimeric sequences, whereby gene sequences from different individuals were concatenated, given that a full complement of genes was not available for all individuals of a species. However, this affected mainly the outgroup with only a few Python species being chimeric (Table 1). Using the BLAST tool to verify GenBank data (https://blast.ncbi.nlm.nih.gov/Blast.cgi), and by checking linked museum records online, some publicly available Python sequences appear to be incorrectly annotated and were therefore not included (GenBank Accessions: MK194022, EF545065, EF545064).

To generate the new sequences, total genomic DNA was extracted from tissue samples preserved in 99% ethanol, using a salt extraction protocol (Aljanabi and Martinez 1997). Gene fragments were amplified with polymerase chain reaction (PCR): mitochondrial ribosomal large subunit (16S 493 bp), cytochrome-b (Cyt-b 1,110 bp), and nuclear (nDNA) gene oocyte maturation factor Mos (c-mos 509 bp) using primers 16S: 16Sa/16Sb (Palumbi et al. 2002); Cyt-b: L14910/H16064 (Burbrink et al. 2000); c-mos: FUF/FUR (Gamble et al. 2008) or S77/S78 (Lawson et al. 2005). Amplification took place using a 25 µl reaction mixture consisting of 2.5 µl reaction buffer, 2.5 mM MgCl₂, 2 µM of each primer, 0.2 mM DNTP solution, 0.5 U/µl Taq Polymerase (SuperTherm), and 25–50 ng/µl of DNA template. The thermal cycling consisted of initial denaturation for 4 min at 95 °C, followed by 35 cycles of denaturation (94 °C, 45 s), annealing (50–60 °C, 30 s), and extension (72 °C, 1 min), with a final extension at 72 °C for 10 min. PCR products were visualised through electrophoresis on a 0.8% agarose gel and Sanger sequencing was carried out at Macrogen (Amsterdam, Netherlands) using the forward primers for each gene. Sequences were aligned in Geneious v.11 (https://www.geneious.com) using MUSCLE alignment, and the coding genes were checked for the presence of stop codons or frame shifts in the codon sequences.

Both Bayesian (MrBayes and BEAST) and the likelihood analyses produced the same topology for Python, with well-supported nodes for each species in the genus (Suppl. material 1: fig. S1). The *BEAST species tree produced the same topology and support as the Bayesian and likelihood analyses (Suppl. material 1: fig. S2). All four African species (P. anchietae, P. natalensis, P. regius, P. sebae) were monophyletic to the exclusion of the four Asian species included in the analysis (P. bivittatus, P. brongersmai, P. curtus, P. molurus). Within the African clade, P. regius is sister the other three African species, having diverged earliest. This placement agrees with that of Reynolds et al. (2014) but contrasts with Pyron et al. (2013) who found P. regius and P. sebae to be within the Asian clade rendering both the African and Asian clades polyphyletic.

The dated phylogeny suggests that the genus Python diverged from other Pythonidae genera around 43 Mya (Fig. 4, Suppl. material 1: fig. S3, table S4). The African and Asian Python clades diverged approximately 33 Mya during the Late Eocene/Early Oligocene (Fig. 3). Most species level diversity in the Asian clade originated in the Late Miocene/Early Pliocene between 15 and 5 Mya (Fig. 4, Suppl. material 1: fig. S3, table S4). Two Asian species are missing from our phylogeny (P. breitensteini and P. kyaiktiyo). It should be noted that P. breitensteini has been shown to be sister to P. curtus (Keogh et al. 2001), but those comparative data were unavailable from GenBank. However, we assume that P. breitensteini – P. curtus divergence also occurred within this same time frame and would be more recent than the split between P. curtus and P. brongersmai (i.e., less than ~13 Mya). Species level divergences within the African clade are generally older, with P. regius diverging around 26 Mya and P. anchietae at about 22 Mya (Fig. 4, Suppl. material 1: table S4). Although P. natalensis was formerly regarded as a subspecies of P. sebae, these taxa are clearly distinct sister species having diverged approximately 12 Mya. Furthermore, the networks showed clear divergence, with no haplotype or allele sharing between P. natalensis and P. sebae for either mitochondrial or nuclear genes, even for the slow-evolving c-mos gene, despite some individuals being sampled from the zones of sympatry (Fig. 5). We acknowledge, however, that to investigate potential gene flow leading to hybridisation or introgression between these species would require denser sampling in the presumed sympatric/parapatric zones.

The BBM analysis showed a high probability for the Oriental biogeographic region as the ancestral area for the genus Python (Fig. 6). The reconstruction also suggests that Python colonised Africa from Asia (Oriental region) around 33 Mya given the divergence time of the African and Asian clades (Fig. 6, node 6 in Suppl. material 1: table S4). However, the presence of Python fossils in the Palearctic (Europe) and the Saharo-Arabian biogeographic regions from the Miocene (Fig. 2) more accurately indicates that Python would have arrived in Africa via those regions rather than directly from the Oriental region.

Figure 4. 

Dated phylogeny for Pythonidae with near-complete taxon sampling for the genus Python. Nodes with black dots supported by all analyses, whereas white dots indicate nodes supported by two of three analyses (Suppl. material 1: figs S1, S3). Body sizes are noted for species or genera (Suppl. material 1: table S5), and example photos of the African species are shown. Bottom: Time scale with geological Epochs labelled. Trend line shows the deep-sea oxygen isotopic (δ ¹⁸O) record as a proxy for (global) temperature since the Eocene (graph reproduced from Zachos et al. 2001a).

Figure 5. 

Network of haplotypes (Top: Cyt-b) and alleles (Bottom: c-mos) for Python natalensis and Python sebae. The size of the circles is proportional to the frequency of individuals with that haplotype/allele, and the branch lengths are proportional to the number of mutations. The branches interrupted by hatch marks are shortened, with the number of mutations along that branch indicated.

Figure 6. 

Probabilities of ancestral ranges at each node in the Python dated phylogeny, represented by pie charts that are colour-coded to the areas as indicated in the key (bottom left). Biogeographic regions (Holt et al. 2013) used to code the terminal taxa (bottom right), with Africa sub-partitioned into Central (C), East (E) and southern (S). Coding used at each terminal is shown by the coloured block(s) to the right of the species name. The relative locations of Africa, Arabia, Asia and Europe at four different time points since the Eocene are depicted (source: C.R. Scotese, PaleoMap Project 2013).

Previous interpretations of the biogeographic history of pythons focussed on Australasian taxa, with limited taxon sampling and geographic coverage of African Python (Keogh et al. 2001; Reynolds et al. 2014; Esquerré et al. 2020, 2022). With improved taxonomic and geographic sampling, we show that all African species are monophyletic to the exclusion of Asian species, a finding that has not been shown previously. Furthermore, the clear divergence between P. natalensis and P. sebae dated at ca. 12 Mya, and the lack of allele or haplotypes sharing despite some individuals being sampled from zones of sympatry, provides strong support for P. natalensis being considered a full species as argued by Broadley (1999). In addition, our results support an Asian origin for the genus, with a single dispersal event into Africa in the early Oligocene (ca. 33 Mya). This was followed by diversification on the African continent beginning around 26 Mya, a time span that was relatively warm compared to the subsequent Miocene cooling (see Fig. 4 – bottom; Westerhold et al. 2020; Couvreur et al. 2021). Species level diversification ended in Africa around 12 Mya concomitant with the acceleration of the mid-Miocene cooling trend (Zachos et al. 2001a, b; Westerhold et al. 2020), and the fossil evidence shows an overall range contraction with some populations and species becoming extinct during that epoch (Fig. 2).

From the high-level dated phylogeny (Fig. 4), it can be inferred that the genus Python diverged from other genera of family Pythonidae in the late Eocene, approximately 43 Mya in Asia (see also Esquerré et al. 2020). Subsequently, the African and Asian clades of Python diverged (ca. 33 Mya) during the Eocene–Oligocene transition. The time frame is concurrent with the Afro-Arabian tectonic plates being partly submerged and positioned such that they were isolated from the Eurasian landmass by the Tethys Sea during the Cretaceous and into the early Cenozoic (Guiraud et al. 2005). Prior to that time, faunal or floral interchange between Afro-Arabia and Eurasia was probably rare and only possible via overseas dispersal, although this may have become more achievable by the early Oligocene as the continental plates drew closer and the marine transgressions receded (Godinot et al. 2003; Seiffert 2012; Hamon et al. 2013). Connections between the Afro-Arabian plates with Eurasia in the Oligocene presumably allowed for faunal and floral interchange (Vermeij 1991; e.g., vipers and cobras – Pook et al. 2009, Varanus – Portik and Papenfuss 2012; primates – Springer et al. 2012), providing an opportunity for at least one lineage of Python to enter the Afro-Arabian landmass.

Contraction of the tropics towards the equator intensified around the Eocene-Oligocene transition (Couvreur et al. 2021), approximately 33 Mya, forcing the African and Asian Python clades to retreat away from each other as the climate became cooler. Currently, environmental temperatures across much of the distribution gap regularly fall below 20 °C (Suppl. material 1: fig. S4), which is a critical threshold for egg incubation in Python (Alexander 2007) and is possibly similar for all Pythonidae. Furthermore, fossil evidence shows that the Arabian Peninsula and the Namib region were occupied by pythons during the early and mid-Miocene (Fig. 2, Suppl. material 1: table S1), prior to the Late Miocene progression toward hyper-aridity in these regions (Sepulchre et al. 2006; Saarinen et al. 2020; Siesser 2020; Couvreur et al. 2021; Doman and Early 2022), suggesting that hyper-aridity is an additional limiting factor. Overall, the present-day distribution of Python can be attributed to life-history traits that limit them to warm environments, ultimately forcing them to contract toward the topics and sub-tropics (Alexander 2007; Suppl. material 1: fig. S4).

The absence of Python on Madagascar both at present and in the fossil record is perhaps unsurprising. Given the post-Gondwanan age of Pythonidae, a Gondwanan distribution – which would include Madagascar – can be ruled out. Furthermore, most Malagasy reptile fauna colonised the island through overseas dispersal from eastern Africa during the Eocene and early Oligocene (ca. 55–35 Mya; Ali and Hedges 2023), a time period prior to the arrival of Python in Africa. It is presumed that paleo-currents were optimal during that period and facilitated colonisation of the island via overseas dispersal from eastern Africa (Ali and Huber 2010; Samonds et al. 2012, 2013; Tolley et al. 2013). The arrival of Python in northern Africa from Asia coincided with the shifting of the relative positions of Africa and Madagascar that changed the relative direction of the oceanic currents away from Madagascar (Ali and Huber 2010), making overseas dispersal to Madagascar much less likely. Thus, the arrival of Python in Africa after the optimal dispersal period toward Madagascar provides a reasonable explanation as to the absence of pythons on the island. Notably, this explanation is independent of climatic drivers that have been invoked to explain the other distribution gaps for Python.

Of the ten species in the genus Python, four qualify as being amongst the largest extant snakes on Earth (Murphy and Henderson 1997; Suppl. material 1: table S5), and all these large-bodied species are limited to the tropical and sub-tropical regions of Asia and Africa (O’Shea 2007). Fossil evidence also indicates that species which previously occurred in the geographic gap between current day African and Asian Python were large-bodied with vertebrae easily exceeding 10 mm in breadth (see Rage and Bailon 2005; Ivanov and Böhme 2011; Georgalis et al. 2020) correlating to body sizes in excess of 4 m (Rage 1976; compare to Head et al. 2009). Given their fossil record, it is likely that the ancestral lineage of Python that extended its range into Africa was large-bodied. Functional traits of snakes such as their elongated shape, ectothermic physiology and prey size selection, which are associated with gape-limitation, are thought to have a significant influence on the evolution of snake body size (Boback and Guyer 2003). For example, it has been suggested that prey size is an important determinant of snake body size, where availability of large prey drives the evolution toward larger body and gape size (Boback and Guyer 2003) so that large-bodied snakes can meet their energy budgets. An alternative hypothesis for the restriction of large snakes to warm areas is that large-bodied snakes warm more slowly than small-bodied snakes while basking, restricting the time spent at target body temperatures in cool environments (Bogert 1949; Alexander 2007). Shine (1994) has also proposed that selection for large body size can result from larger females due to fecundity advantages. Thus, the evolution of large body size may be linked to warm climates, abundance of mega-faunal prey which existed on both continents (Steinthorsdottir et al. 2021) and/or reproductive advantages linked to large body size.

While some areas of Africa are hyper-diverse biologically (e.g., Fjeldså and Lovett 1997; Bayliss et al. 2024), Africa stands out as having relatively low species richness compared to other tropical regions (Couvreur 2015; Raven et al. 2020; Hagen et al. 2023), consistent with the low Python richness in Africa. Along these lines, the adaptive diversification that occurred with the invasion of Australasia by Pythonidae (27 species from 9 genera; Esquerré et al. 2017, 2022) was far more dramatic than for Africa (4 species from one genus). The high richness for pythons in Australasia has been explained by the colonization of environmentally diverse regions with novel niches, promoting allopatric speciation (Esquerré et al. 2022), consistent with the hypothesis that speciation rates are higher outside Africa. The comparatively low speciation rates for Africa appear to be driven by the unique combination of an arid climate, reduced topographic and habitat complexity, and long-term geographic isolation through the early Cenozoic (Couvreur 2015; Raven et al. 2020; Couvreur et al. 2021; Hagen et al. 2023). These same environmental features probably also reduced opportunities for Python to speciate in Africa, limiting niche diversity and reducing allopatric speciation due to an arid, homogeneous habitat.

With improved taxon sampling compared to previous studies, we showed that the most likely geographic origin of genus Python is Asia, with a single dispersal event into Africa corresponding with the closing of the Tethys Sea and newly established contact of the African and Eurasian landmasses. Our interpretation, however, is based on a dataset that includes few loci (two mitochondrial and one nuclear genes) and confidence in this hypothesis could be improved through more detailed genomic datasets and approaches. Nevertheless, our overall topology agrees with recent analyses that incorporate genomic approaches (Esquerré et al. 2020, 2022), thus, we do not expect that our current interpretation, despite being based on limited gene sampling, would notably change.

At present, Python species occur in mesic to humid sub-tropical or tropical regions and are excluded from temperate and arid regions in both Asia and Africa, with the exception of P. anchietae which has adapted to arid (but not hyper-arid) conditions. The fossil record shows that the present-day distribution gap was once occupied by pythons up until the mid-Miocene, during an era when those areas were warmer and more mesic. Thus, the present day and presumed paleo-distributions based on fossil evidence support the hypothesis that climate, particularly temperature and aridity, can be considered strong range limiters for pythons due to decreased recruitment of offspring outside warm and mesic environments.

This project was funded by the South African National Biodiversity Institute. We are grateful to a number of individuals who contributed to the DNA sequencing for this study, including Ninda Batista, Aaron Bauer, Marius Burger, Werner Conradie, Anja le Grange, Javier Lobón-Rovira, and Pedro vaz Pinto, and to Thomas Couvreur, Werner Conradie and Damien Esquerré for their helpful comments and advice. The South African National Wildlife Biobank provided additional samples for sequencing. Photos used in the figures were kindly provided by Bianca Fizzotti, Javier Lobón-Rovira, Michele Menegon and Steve Spawls.

Conceptualization: KAT, GJA. Formal analysis: KAT. Funding acquisition: KAT. Investigation: KAT, GJA. Writing – original draft: KAT, GJA. Writing – review and editing: KAT, GJA.

Genetic data are accessible on GenBank as per Table 1.